Abstract: | This investigation builds upon the work done by Emma Hart, Peter Ross and Jeremy Nelson. Their work focused upon using Evolutionary techniques to solve a real world, hevily constrained scheduling problem for a local chicken factory. Strategies for solving this scheduling problem were addressed implementing two separate, yet intrinsically dependent GAs to produce a candidate solution. While their research produced successful results, several key areas were identified for further investigation.This investigation addresses one such area; namely that of producing multiple and equally viable schedules by optimizing the non-linear variable relationships of a candidate solution through techniques associated with the Parallel Island Genetic Algorithm model. since an Island GA model incorporates pseudo parallel processing, combining these techniques allows for the differentiation of an encoded problem, thus producing several candidate solutions from a single starting chromosome.This investigation concentrates on the qualitative feasibility of using such Island Genetic Algorithm techniques to improve the overall method for producing successful Chicken Catching schedules.
|